You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by Mike Thomsen <mi...@gmail.com> on 2016/11/09 17:17:22 UTC

Rolling backups of a collection

I read over the docs (
https://cwiki.apache.org/confluence/display/solr/Making+and+Restoring+Backups)
and am not quite sure what route to take. My team is looking for a way to
backup the entire index of a SolrCloud collection with regular rotation
similar to the backup option available in a single node deployment.

We have plenty of space in our HDFS cluster. Resources are not an issue in
the least to have a rolling back up of say, the last seven days. Is there a
good way to implement this sort of rolling backup with the APIs or will we
have to roll some of the functionality ourselves?

I'm not averse to using the API to dump a copy of each shard to HDFS.
Something like this:

/solr/collection/replication?command=backup&name=shard_1_1&numberToKeep=7

Is that a viable route to achieve this or do we need to do something else?

Thanks,

Mike

Re: Rolling backups of a collection

Posted by Mike Thomsen <mi...@gmail.com>.
Thanks. If we write such a process, I'll see if I can get permission to
release it. It might be a moot point because I found out we're stuck on
4.10.3 for the time being. Haven't used that version in a while and forgot
it didn't even have the collection backup API.

On Wed, Nov 9, 2016 at 2:18 PM, Hrishikesh Gadre <ga...@gmail.com>
wrote:

> Hi Mike,
>
> I filed SOLR-9744 <https://issues.apache.org/jira/browse/SOLR-9744> to
> track this work. Please comment on this jira if you have any suggestions.
>
> Thanks
> Hrishikesh
>
>
> On Wed, Nov 9, 2016 at 11:07 AM, Hrishikesh Gadre <ga...@gmail.com>
> wrote:
>
> > Hi Mike,
> >
> > Currently we don't have capability to take rolling backups for the Solr
> > collections. I think it should be fairly straightforward to write a
> script
> > that implements this functionality outside of Solr. If you post that
> > script, may be we can even ship it as part of Solr itself (for the
> benefit
> > of the community).
> >
> > Thanks
> > Hrishikesh
> >
> >
> >
> > On Wed, Nov 9, 2016 at 9:17 AM, Mike Thomsen <mi...@gmail.com>
> > wrote:
> >
> >> I read over the docs (
> >> https://cwiki.apache.org/confluence/display/solr/Making+and+
> >> Restoring+Backups)
> >> and am not quite sure what route to take. My team is looking for a way
> to
> >> backup the entire index of a SolrCloud collection with regular rotation
> >> similar to the backup option available in a single node deployment.
> >>
> >> We have plenty of space in our HDFS cluster. Resources are not an issue
> in
> >> the least to have a rolling back up of say, the last seven days. Is
> there
> >> a
> >> good way to implement this sort of rolling backup with the APIs or will
> we
> >> have to roll some of the functionality ourselves?
> >>
> >> I'm not averse to using the API to dump a copy of each shard to HDFS.
> >> Something like this:
> >>
> >> /solr/collection/replication?command=backup&name=shard_1_1&
> numberToKeep=7
> >>
> >> Is that a viable route to achieve this or do we need to do something
> else?
> >>
> >> Thanks,
> >>
> >> Mike
> >>
> >
> >
>

Re: Rolling backups of a collection

Posted by Hrishikesh Gadre <ga...@gmail.com>.
Hi Mike,

I filed SOLR-9744 <https://issues.apache.org/jira/browse/SOLR-9744> to
track this work. Please comment on this jira if you have any suggestions.

Thanks
Hrishikesh


On Wed, Nov 9, 2016 at 11:07 AM, Hrishikesh Gadre <ga...@gmail.com>
wrote:

> Hi Mike,
>
> Currently we don't have capability to take rolling backups for the Solr
> collections. I think it should be fairly straightforward to write a script
> that implements this functionality outside of Solr. If you post that
> script, may be we can even ship it as part of Solr itself (for the benefit
> of the community).
>
> Thanks
> Hrishikesh
>
>
>
> On Wed, Nov 9, 2016 at 9:17 AM, Mike Thomsen <mi...@gmail.com>
> wrote:
>
>> I read over the docs (
>> https://cwiki.apache.org/confluence/display/solr/Making+and+
>> Restoring+Backups)
>> and am not quite sure what route to take. My team is looking for a way to
>> backup the entire index of a SolrCloud collection with regular rotation
>> similar to the backup option available in a single node deployment.
>>
>> We have plenty of space in our HDFS cluster. Resources are not an issue in
>> the least to have a rolling back up of say, the last seven days. Is there
>> a
>> good way to implement this sort of rolling backup with the APIs or will we
>> have to roll some of the functionality ourselves?
>>
>> I'm not averse to using the API to dump a copy of each shard to HDFS.
>> Something like this:
>>
>> /solr/collection/replication?command=backup&name=shard_1_1&numberToKeep=7
>>
>> Is that a viable route to achieve this or do we need to do something else?
>>
>> Thanks,
>>
>> Mike
>>
>
>

Re: Rolling backups of a collection

Posted by Hrishikesh Gadre <ga...@gmail.com>.
Hi Mike,

Currently we don't have capability to take rolling backups for the Solr
collections. I think it should be fairly straightforward to write a script
that implements this functionality outside of Solr. If you post that
script, may be we can even ship it as part of Solr itself (for the benefit
of the community).

Thanks
Hrishikesh



On Wed, Nov 9, 2016 at 9:17 AM, Mike Thomsen <mi...@gmail.com> wrote:

> I read over the docs (
> https://cwiki.apache.org/confluence/display/solr/
> Making+and+Restoring+Backups)
> and am not quite sure what route to take. My team is looking for a way to
> backup the entire index of a SolrCloud collection with regular rotation
> similar to the backup option available in a single node deployment.
>
> We have plenty of space in our HDFS cluster. Resources are not an issue in
> the least to have a rolling back up of say, the last seven days. Is there a
> good way to implement this sort of rolling backup with the APIs or will we
> have to roll some of the functionality ourselves?
>
> I'm not averse to using the API to dump a copy of each shard to HDFS.
> Something like this:
>
> /solr/collection/replication?command=backup&name=shard_1_1&numberToKeep=7
>
> Is that a viable route to achieve this or do we need to do something else?
>
> Thanks,
>
> Mike
>